Greyline Charts, SNR v Signal and noise.

I have many recordings over the years..

The Format of Charts I Produce.

The charts show the signal strength as received. The receiver is set to CW or SSB mode to produce a heterodyne with the signal's carrier. Spectrum Laboratory (SpecLab) is set to export a text file to record the amplitude of this audio signal. The receiver is operated with no AGC, provided it is not overloaded the audio volume represents the actual signal strength received. SpecLab produces a waterfall, showing signal strength as a change in brightness or colour, see below. A chart shows this more clearly.

These charts are produced with Excel. Open Office or SpecLab's "Watch and Plot" may be used. "Watch and Plot" charts may be saved by a "Scheduled Action" eg. plot.capture("C:\SpecLab2\5045\5045Chart-"+str("DD-MM-YY",now)+".jpg") Look *** for more information.

The top, purple line represents the signal strength of Macapa, the station being observed, gradually rising until it is at it's strongest just before the yellow line showing sunrise. Do not take this as a typical example, if you look at my other charts you will see great variations in propagation patterns. Indeed prolonged observations make me think that if there are propagation patterns on lower HF frequencies they may take years to determine.

I have complicated this chart by recording a second signal seen about 30Hz away from Macapa. The blue trace shows it has a similar sunset time and that it switches off at 22.40. The blue trace follows the noise level when nothing is received.

The black line at the bottom represents my local background noise level, giving an indication of the signal to noise ratio. Sometimes this may rise with interference and swamp the signal showing that the recording at that time is unreliable.

The green line represents the received frequency of the signal. It is adjusted so it fits into the chart with the dB scale giving an indication of variation in Hz. For example the above chart shows that at about 23.00 the frequency begins to rise about 7Hz, dropping 3Hz then gradually rising again. This is shown on the "waterfall", see below, the frequency scale here shows the actual received frequency before I subtracted about 3160Hz so it fitted the chart. Various effects affect the frequency recorded, a jump often indicates that there is a stronger signal on a slightly different frequency that is disturbing the recording. My illustration appears to be due to the transmitter varying, (maybe a change of transmitter?) a slight break in transmission occurs immediately before. Receiver or propagation effects also cause variations. Sometimes frequency variations can help identify the station. Some show a zig-zag line, probably due to the crystal oven, this pattern can be confirmed by another SWL who can actually hear the station. ZL1BPU helped me confirm ZLXA by simultaneously noting frequency jumps of this station.

This screen capture shows Macapa between 3020-3030Hz and the other signal at about 3060Hz. These are easily separated into separate lines on SpecLabs's text export. The receiver is a Softrock, using SpecLab sampling at 8KHz to minimise CPU usage with the local oscillator approximately 3KHz below the signal . A conventional receiver using CW mode will produce around 800Hz +/- half the filter bandwidth, SSB can be used, the actual audio frequency is not critical, SpecLab export is just set to it.

Spectrum Laboratory is a versatile tool. It does take time to learn to use. It will produce its own charts. The one below shows 24 hours of two medium wave signals.

SNR figures

Signal to Noise, WSPR users note!

For true analysis of propagation signal AND noise need to be measured at both ends of the path. SNR figures do not show why differences occur.

Note nearly all my observations are in the lower HF bands. I have looked at the WARC beacons for signs of enhancement but only briefly.

Especially on the lower frequencies background noise can vary significantly. Sometimes due to locally generated noise, sometimes due to atmospheric noise received from lightning discharges from considerable distances.

The chart below shows reception of the time signal from Equador on 3.810MHz so will demonstrate 80m propagation. The top purple line represents signal strength. The black line shows background noise level and the blue line shows the difference between these two, the signal to noise ratio.

Note the dB scale is negative so the strongest signal, the greater value is at the top.

The purple line shows that signal strength peaks around 03.00 UTC then stays just 2-3dB below that level until just after sunrise at 05.49.

The black noise line starts to drop before sunrise so that the received signal to noise ratio actually peaks around 06.15. I think this is what WSPR will report, but the best S/N does not mean a sunrise peak!

The chart below, from another day, does show a small peak at sunrise. Combined with the drop in noise level reception quality peaks. Whether this time is best for communication will depend upon the noise level at the other end, expect it to be very high in the tropics!

I have been studying sunrise/sunset enhancements for several years. So far I have found only a few paths which show significant peaks, those that do show no consistency. Many have witnessed this effect but I do not think it is reliable! Alan G4ZFQ

December 2010.

Charts produced using Spectrum Laboratory. http://www.qsl.net/dl4yhf/spectra1.html

Some more information about my use of SpecLab. http://homepages.wightcable.net/~g4zfq/SpecLab.htm This page shows a significant dawn enhancement on 3.810MHz but this is NOT typical!

What is a Peak?

If sunrise and sunset enhancements are seen when do we consider them significant in terms of radio communication? The chart below illustrates the problem. The red trace shows reception of a Brazilian broadcast as received in England.

This chart shows definite peaks which certainly seem to be associated with sunset and sunrise.

If the period around 21.20 were observed in isolation then it could be concluded that a peak occurred. However, after three hours signals become significantly stronger. Sunset may well NOT be the best time for communication.

Can we consider the peak at 08.00? Signals at around 04.00 were at least as strong for a longer period. Also the background noise is slightly lower. The chart only shows the situation at my location but it may be concluded that the best time for communication could be around 04.00.

In fact this chart is not representative of the characteristics of this path. It is unusual to see a peak around sunset or sunrise that reaches the strength of signals during the night-time, let alone exceed them.

In the five years I have been looking for significant sunrise or sunset enhancements I am coming to the conclusion that these are rare. Many UK amateurs operating on 80 metres will have witnessed the way a peak can follow sunrise across Australia. First East coast stations peak, then those more to the West. BUT how often does this happen?

WSPR-SNR-Plot-v-WWCR An attempt to compare WSPR SNR figures.

Unfinished https://sites.google.com/site/g4zfqradio/greyline_charts_using_speclab migrating from here ~g4zfq/SpecLab.htm

Google Greyline Pages A few pages showing "research" using Softrocks.

http://sites.google.com/site/g4zfqradio/3250-1109

http://sites.google.com/site/g4zfqradio/3250-1209

http://sites.google.com/site/g4zfqradio/3810-jan-feb-2009

http://sites.google.com/site/g4zfqradio/signal-to-noise-wspr-users-note

http://sites.google.com/site/g4zfqradio/what-is-a-peak

November 2020, Still monitoring but the number of suitable signals is limited. I am active on WSPR but as stated above it is not a particularly good tool. There is a version that reports the actual noise figure but that needs a little knowledge.